CVR (Conversion Rate)

Key Performance Indicator (KPI) that measures the proportion of visitors who completed an expected action on a site or application, relative to the total number of visitors exposed.
It is expressed as a percentage and is calculated using the formula:

CVR (%) = (Number of conversions ÷ Number of visitors) × 100

🎯 Why it's a central KPI:

CVR is the main evaluation criterion in most A/B or MVT tests, as it reflects the ability of a page, message, or journey to turn a visit into real value: purchase, lead, sign-up, strategic click, etc.

📌 Examples of actions counted as a conversion:

• E-commerce: completed order, add to cart, stock alert sign-up,
• Lead gen / SaaS: form submission, demo request, account creation,
• Content: guide download, newsletter sign-up, complete video viewing.

🔍 Role of CVR in A/B testing:

When testing a variation, the goal is generally to increase the CVR compared to the control version. The test makes it possible to determine, with a certain level of statistical confidence, whether the change made has a positive, neutral, or negative causal impact on this KPI.

CVR is therefore the primary KPI most often used to:

• evaluate the performance of a single variation (A/B test),
• compare several combinations of changes (MVT),
• measure the impact of dynamic or targeted personalizations.

🧪 CVR & test structure:

• In an A/B test, we measure the CVR of the variation vs. that of the control to isolate the effect of a change.
• In an MVT test, each combination has its own CVR, and we analyze the individual or cross-impact of multiple elements (title + image + CTA, etc.).
• The CVR can be “leading” or “lagging” depending on the nature of the conversion:
  • Leading (e.g., clicking a button) → fast-to-measure KPI, useful for detecting early signals,
  • Lagging (e.g., final purchase) → strategic KPI, but longer to observe.

✅ Best practices:

• Always combine CVR with other contextual metrics (e.g., bounce rate, revenue per visitor, segmentation by device or source),
• Check the statistical significance of the result before drawing a conclusion,
• Take into account the Minimum Detectable Effect (MDE) to determine the minimum test duration,
• Cross-reference CVR with quality indicators (e.g., product return rate, unsubscribe rate if CVR is lead-based).

📈 CVR ≠ absolute success:

A good conversion rate is only relevant if it is aligned with the business objective.
A very high CVR on an intermediate action (e.g., clicking “start now”) can mask disappointing performance on the final action (e.g., paid subscription).
This is why CVR must be integrated into an overall reading of the funnel.